Monday, January 11, 2021

The brain: Different expert systems propose strategies for action, keeping track of the precision of the predictions within each system, exerting control over many different expert systems simultaneously to produce sophisticated behavior

Why and how the brain weights contributions from a mixture of experts. John P. O’Doherty et al. Neuroscience & Biobehavioral Reviews, January 11 2021. https://doi.org/10.1016/j.neubiorev.2020.10.022

Rolf Degen's take: https://twitter.com/DegenRolf/status/1348683591333670914

Highlights

• The brain can be thought of as a “Mixture of Experts” in which different expert systems propose strategies for action.

• This is accomplished by keeping track of the precision of the predictions within each system, and by allocating control over behavior in a manner that depends on the relative reliability of those predictions.

• This reliability-based control mechanism is domain general, exerting control over many different expert systems simultaneously in order to produce sophisticated behavior.

Abstract: It has long been suggested that human behavior reflects the contributions of multiple systems that cooperate or compete for behavioral control. Here we propose that the brain acts as a “Mixture of Experts” in which different expert systems propose strategies for action. It will be argued that the brain determines which experts should control behavior at any one moment in time by keeping track of the reliability of the predictions within each system, and by allocating control over behavior in a manner that depends on the relative reliabilities across experts. fMRI and neurostimulation studies suggest a specific contribution of the anterior prefrontal cortex in this process. Further, such a mechanism also takes into consideration the complexity of the expert, favoring simpler over more cognitively complex experts. Results from the study of different expert systems in both experiential and social learning domains hint at the possibility that this reliability-based control mechanism is domain general, exerting control over many different expert systems simultaneously in order to produce sophisticated behavior.

Keywords: cognitive controlPrefrontal cortexbasal gangliaTheoretical neuroscienceDecision-making


Conclusion

Here we outline a framework for conceptualizing the contribution of multiple systems to behavioral control in the human brain. We suggest that the brain utilizes a framework loosely analogous to the mixture of experts in machine learning, in which a prefrontal-based manager (which we hypothesize specifically involves the ventrolateral prefrontal cortex), reads out the reliability of the predictions by each of the constituent experts, and uses these predictions to allocate control over behavior to the experts in a manner that is proportional to the relative precision or uncertainties in their predictions. We suggest that this reliability-based arbitration process between experts is both necessary and sufficient for the efficient allocation of control between systems, as this approach takes into account not only the accuracy and hence the average expected value of the actions nominated by each expert, but also implicitly takes into account the cognitive costs and cognitive constraints. The interaction between systems that makes up the experts is we suggest, better conceived of as one of polling the advice from different systems that each have different relevant expertise that can and should be respected owing to differences in the nature of the information that is being processed, and in the algorithmic transformations that are performed on that information. These experts should be listened to as a collective, because they provide the right mixture of opinions needed to act in the world effectively.

Our results suggest that the theory cannot fully explain human aesthetic responses to flowers and clear preferences for them

Hula, M., & Flegr, J. (2021). Habitat Selection and Human Aesthetic Responses to Flowers. Evolutionary Human Sciences, 1-49. doi:10.1017/ehs.2020.66

Abstract: Although the aesthetic appreciation of flowers is a well-known aspect of human behaviour, theories explaining its origin are missing. The only exception is the evolutionary theory of Heerwagen and Orians. Surprisingly, it has not yet been empirically tested. The authors suggest that humans aesthetically respond to flowers because they signal food availability. The logic of the theory implies that fruits are more reliable and direct food availability signals than flowers. Therefore, fruits should elicit stronger aesthetic responses than flowers. To test this assumption, we performed two online studies in the Czech Republic. The participants (n = 2792 and 744 respectively) indicated on a six-point scale their aesthetic response to photographs of 14 edible Czech plant species (study A) and 20 edible plant species from African savannas (study B), varying in growth stage (flowering, fruiting). We found no difference between the Czech fruiting and flowering plants and a stronger aesthetic response to African flowering plants. A third study (n = 816) confirmed that flowers were preferred to fruits, using a forced-choice paradigm. Our results suggest that the theory cannot fully explain human aesthetic responses to flowers. We discuss alternative explanations. This topic deserves renewed attention from researchers working in related fields.

Social Media summary: Contrary to the assumptions of the habitat selection theory, flowers elicit stronger aesthetic responses than fruits.

Keywords: evolutionary aesthetics, habitat selection, flower preference, perception of flowers


Sexual prejudice toward gays is rooted in working-class experiences; contrary to mainstream ideas, social class matters in contemporary society, the relationship is not spurious, & education is not the main issue

Class Foundations of Sexual Prejudice toward Gay and Lesbian People. Stef Adriaenssens, Jef Hendrickx & Johanna Holm. Sexuality Research and Social Policy, Jan 11 2021. https://rd.springer.com/article/10.1007/s13178-020-00525-y


Abstract

Introduction: Sexual prejudice negatively affects the quality of life and life chances of those involved. Manual workers are consistently found to be less accepting of homosexuality in studies of sexual conformism. This can be seen as an application of Lipset’s ‘working class conformism’. Our core hypothesis is that this lower tolerance is rooted in working-class experiences. Counter-arguments are that that social class does not matter in contemporary society and that the relationship is spurious, with education as the true cause.

Methods: We test the central hypothesis with European survey data. First, we regress sexual prejudice on time trends and class with repeated cross-sections from the European Social Survey, ranging from 2002 to 2016. As an extra check, this is also applied to the European Values Study, going back to 1981. Further, we test the spuriousness argument with a matching design, testing whether stratification accounts for the lag.

Results: The time series shows a stable lag between working-class members and others against the general trend of decreasing sexual prejudice. The matching design provides evidence that working-class membership in itself is a factor behind differences in sexual prejudice.

Conclusions: Contrary to ‘death of class’ conjectures, working-class membership is related to sexual prejudice. This contribution shows that this gap is due to experiences of belonging to the working class and not solely to educational differences.

Policy Implications: Occupational experiences, especially in low-skill manual labour, have social effects in areas such as sexual prejudice. Improving the quality of work thus facilitates a more inclusive society for sexual minorities.


Chimpanzees stopped bartering when the human no longer mediated the trade; the authors think that the issue holding them back was an inability to trust their partners & a lack of third party enforcement mechanisms

What behaviour in economic games tells us about the evolution of non-human species' economic decision-making behaviour. Sarah F. Brosnan. Philosophical Transactions of the Royal Society B: Biological Sciences, Volume 376, Issue 1819, January 11 2021. https://doi.org/10.1098/rstb.2019.0670

Rolf Degen's take: https://twitter.com/DegenRolf/status/1348531514955460609

Abstract: In the past decade, there has been a surge of interest in using games derived from experimental economics to test decision-making behaviour across species. In most cases, researchers are using the games as a tool, for instance, to understand what factors influence decision-making, how decision-making differs across species or contexts, or to ask broader questions about species’ propensities to cooperate or compete. These games have been quite successful in this regard. To what degree, however, do these games tap into species' economic decision-making? For the purpose of understanding the evolution of economic systems in humans, this is the key question. To study this, we can break economic decision-making down into smaller components, each of which is a potential step in the evolution of human economic behaviour. We can then use data from economic games, which are simplified, highly structured models of decision-making and therefore ideal for the comparative approach, to directly compare these components across species and contexts, as well as in relation to more naturalistic behaviours, to better understand the evolution of economic behaviour and the social and ecological contexts that influenced it. The comparative approach has successfully informed us about the evolution of other complex traits, such as language and morality, and should help us more deeply understand why and how human economic systems evolved.

Check also Acquisition of object-robbing behavior in macaques: After stealing inedible & more or less valuable objects from humans, they appear to use them as tokens, by returning them to humans in exchange for food

Acquisition of object-robbing and object/food-bartering behaviours: a culturally maintained token economy in free-ranging long-tailed macaques. Jean-Baptiste Leca, Noƫlle Gunst, Matthew Gardiner and I. Nengah Wandia. Philosophical Transactions of the Royal Society B: Biological Sciences, Volume 376, Issue 1819, January 11 2021. https://www.bipartisanalliance.com/2021/01/acquisition-of-object-robbing-behavior.html


4. What can we say about the evolution of economic systems?

Economic games are a model system that allows for close comparisons across contexts and species [46]. However, they are not intended to be natural situations, thus the data generated are best used to test hypotheses generated from observational or experimental work and generate new hypotheses that may be tested with more species-specific and/or naturalistic approaches [8]. While this can be done in diverse ways, there are two literatures that obviously connect to the existing data on experimental economics, that on cooperation and that on trade. Both literatures are extensive enough for entire reviews on their own, but I briefly discuss them as they tie into the main points raised above.

The assurance game highlights the fact that not all species coordinate, and even among those that do, coordination may be achieved by different underlying mechanisms. This reflects their behaviour in natural contexts as well. For instance, in capuchins, group territorial defence (i.e. [47]) and coordinated defence against predators (i.e. snake mobbing: [48]) are typically done among individuals with a clear view of one another and in situations in which all individuals are performing similar actions. This is reminiscent of capuchins’ behaviour in the Assurance game, in which they appeared to coordinate by matching their partners, and only achieved coordination when they could see the other's outcomes. This suggests that behaviour matching is a key mechanism for coordination in capuchins, possibly facilitated by behavioural synchrony and/or social facilitation. Indeed, although capuchins in experimental contexts can solve coordination tasks that require different actions (i.e. [41]), this was a rather simple case in which individuals could see one another and performed actions that they previously learned individually as a sequence.

On the other hand, chimpanzees show more complex coordinated behaviour, such as the coordinated hunting in which different individuals take on different roles, seen in some populations [49], which presumably requires an ability to understand the larger picture. This, too, matches their behaviour in the economic games, in which at least some chimpanzees (those with extensive experience with cognitive tasks) extrapolated to novel options in coordination games, suggesting that they understood their choice as part of a larger strategy. In some experimental contexts, too, chimpanzees show evidence of understanding the bigger picture, for instance choosing to benefit, at a cost to themselves, a chimpanzee that previously paid a cost to make that choice available [44]. Such specificity in how they make decisions suggests that they see their choice within a broader framework.

This synchrony between the outcomes of the highly structured economic games and these primates' natural behaviours suggests that the games are indeed useful models. They allow for studying mechanisms that may be impossible to study in more natural situations and for comparison across populations or species, and may be particularly helpful if differences in body form or ecology make it impossible to compare in more naturalistic contexts. Within the same population or species, these games can also be used to generate new hypotheses regarding behaviour, and the mechanisms underlying it, in more naturalistic contexts. Future work using more naturalistic experimental tasks, like the cooperative barpull (reviewed in [50]), or field-based experiments (as has been done in social learning research; [51]) that combine the naturalistic contexts of the observational work with the structured methodology of the economic games will help determine the limits and flexibility of the different species’ coordination ability and the cognitive mechanisms that underlie them.

We can similarly learn from how primates trade goods and services. Primates' ability to find the NE in economic games suggests that they should be able to maximize their outcomes in barter, too. Indeed, in natural contexts, primates do trade, although typically this involves services (grooming, support in conflicts, and mating opportunities) rather than goods (there are exceptions; for instance, chimpanzees trade meat for mating opportunities and support; [52]). There are several possible reasons for this; objects are zero sum commodities, and few items in primates’ natural lives are worth trading, as most are either easy to acquire or cannot be stored for future trades [5]. This does suggest, however, that they should trade objects when it is worth doing so, and indeed, in experimental tasks, primates trade tokens with experimenters (or, sometimes, other primates) in order to obtain different food rewards (see Addessi et al. [53] and Beran & Parrish [54]). Some macaques have even spontaneously developed exchange systems with humans in free-ranging contexts [55].

What is notably absent, however, is the transfer of these barter relationships with humans to trade with one another. We tested this among three highly trained chimpanzees at Georgia State's Language Research Center (the same chimpanzees that showed evidence of strategy use in the Assurance game). These chimpanzees had been taught a symbol language that allowed us to communicate with more specificity than is typically possible [56]. We made tokens representing specific foods (labelled by their symbol) that they could exchange back to an experimenter for food if—and only if—that food was present in their personal bin. In a series of tasks, we then explored whether the chimpanzees would learn to trade tokens that were of no value to them (because the food was not in their bin) to a partner to whom the tokens were valuable (because the food was in the partner's bin) so as to maximize both chimpanzees' benefits [57].

To cut a long story short, the chimpanzees learned to do so effectively as long as a human experimenter mediated the interactions such that neither chimpanzee could exchange a token with the experimenter for food until they had reciprocated any trades from the partner (we did not restrict which token they had to trade to a partner, just that they traded something). Despite having previously maximized their rewards, within one session of us removing experimenter-mediated quid pro quo, the chimpanzees ceased trading any tokens with their partners. Since they demonstrated all of the necessary cognitive abilities to understand trade, we hypothesized that the issue holding them back was an inability to trust their partners and a lack of third party enforcement mechanisms to make doing so worthwhile [57].

In concert with the results from experimental games, this suggests that at least some primates have the cognitive ability to understand these strategies and maximize their outcomes, which may be evident in service markets, but lack the enforcement mechanisms that lead to beneficial trade of goods. This is important for two reasons. First, it suggests that even these very structured, artificial laboratory situations are indeed useful for understanding the true scope of primates’ abilities, by removing factors that may inhibit expression. Second, this suggests that primates have a more developed set of abilities related to economic behaviour than is necessarily indicated in their natural behaviour. This not only guides future research aimed at understanding economic behaviour in natural contexts, but suggests that humans' abilities are not as separate from the other primates' as we might think. Indeed, this isn't the only context in which experimental tasks have revealed such similarities; primates also share psychological adaptations related to economic decision-making, such as the endowment effect [58] and framing effects [59,60]. Increasing evidence suggests that other species have the cognitive toolkit for economic behaviour and simply lack the opportunity to use it.

If this is the case, our next step is twofold. First, we must explore the range of economic behaviours in other species, ideally with the goal of delineating possible underlying cognitive mechanisms, and determine the contexts in which animals show these behaviours and how they are influenced by changes in social and ecological context. This will require creative experimental studies (both in the laboratory and, hopefully, in the field) to unpack what underlying abilities are present and when they manifest, as well as observational research to determine if related behaviours are seen in more natural contexts [61]. Indeed, further work in this area will also help to clarify the degree to which these abilities were selected specifically for this context versus others. Second, to more fully understand the cognitive precursors to economic behaviour, we also need to look beyond the primates. Although there has been work in this direction already, with studies of economic game behaviour in species as diverse as birds [62], rodents [63] and fish [64,65], even in those cases the focus has been on one or a few species. A more diverse approach is important for understanding how economic behaviour evolved; species vary in their needs, so a broad exploration will inform our understanding of the contexts, environments and pressures that selected for different economic behaviours. Ultimately, this will clarify how these abilities expanded so greatly in humans to result in the complex economic systems we enjoy today.